提升思維層次

導讀

影片連結：

https://www.ixigua.com/6906033244232221191

本影片釋出於2020年12月14日，播放量已超四百萬

本文為袁嵐峰對話諾獎得主約阿希姆·弗蘭克之二，

系列之一見

《袁嵐峰對話諾獎得主約阿希姆·弗蘭克（一）神器冷凍電鏡究竟是什麼？諾獎發明者親自解說 | 科技袁人》

【約阿希姆·弗蘭克與袁嵐峰對話】

袁嵐峰：

You said that there is such a technique that lets the molecules feel at home. So it seems to me that the technique of the super-fast cooling right? The temperature decreases by more than 10,000 degrees in one second. Is that the technique that made the name cryo-EM right?

你說有這樣一種技術，“讓分子有在家的感覺”。所以在我看來這就是超快冷卻技術對嗎？溫度以一秒鐘內一萬多度的速度下降。是這種技術讓它被叫做冷凍電鏡的嗎？

約阿希姆·弗蘭克：

Yes, the molecules are embedded in water. So they’re embedded by very quickly freezing, and when the freezing is fast enough, then the water doesn't have a chance to expand and form crystals.

是的，分子嵌入水中。所以它們是透過很快的凍結來嵌入的，當凍結足夠快時，水就沒有機會膨脹並形成晶體。

And so the molecule is again like in a native environment because the water stays in the same structure like glass, you know, whereas if you cool very slowly, then the water forms crystals, which expand and then they crush the molecule.

因此這個分子感覺像是在它的原生環境中，因為水保持在相同的結構上，就像玻璃一樣。然而如果你冷卻得很慢，水就會形成晶體，晶體會膨脹，然後壓碎分子。

袁嵐峰：

Yeah, because my background is a theoretical and computational chemistry, so I can understand that under such a super-fast freezing the water becomes amorphous ice, not crystal ice, because the hydrogen bonds in the crystal ice have some ice rule, so the density of crystal ice is lower than that of liquid water. And so the volume of a crystal ice is larger than liquid water, so they develop to squeeze the cells and destroy the molecule, is my understanding correct?

是的，因為我的背景是理論與計算化學，所以我能理解在如此快速的冷凍速度下，水變成了無定形冰，而不是晶體冰。因為晶體冰中的氫鍵有一定的冰規則（每個水分子都跟周圍的四個水分子形成氫鍵），晶體冰的密度比液態水低，所以晶體冰的體積比液態水大，因此它們會擠壓細胞，破壞分子，我的理解正確嗎？

約阿希姆·弗蘭克：

Yes, exactly.

是的，非常正確。

袁嵐峰：

So, if you are cooling them in a such a super-fast speed, then you will get amorphous ice, but my question is this, is the density of amorphous ice the same as the liquid water?

所以，如果你以超高速冷卻它們，你會得到無定形冰，但我的問題是，非定形冰的密度和液態水的密度一樣嗎？

約阿希姆·弗蘭克：

I think it's very close.

是的，非常接近。

袁嵐峰：

Okay, I see. So you can explain this phenomenon in terms of density. Okay, this sounds very interesting. So I read some introduction to the contributions of you three, you and Professor Henderson and Professor Dubochet. In my understanding, your contributions are like this, So Professor Henderson found a way to find a structure that if the molecules are aligning in the crystalline way and in this way, it can accept very little radiation, right? so they can tolerate the radiation.

好吧，我明白了。所以你可以用密度來解釋這個現象。好吧，這聽起來很有趣。我讀了一些介紹你們三個，你，亨德森教授和杜波謝教授的貢獻。在我的理解中，你的貢獻是這樣的：亨德森教授找到了一種方法，來確定分子的結構，如果分子以晶體的方式排列，它可以受到更少的輻射，對吧？因此他們能承受輻射。

約阿希姆·弗蘭克：

That's again the same idea to distribute the radiation load among many copies of the molecule.

And in one case, these copies are aligned with one another in an order arrangement and the other case, they are single. But the idea of distributing radiation load is the same, and then the later important step, is averaging, only by averaging, do you get the signal up again.

這和把輻射負載分配到分子的多個複製中的想法是一樣的。

在一種情況下，這些副本按照順序排列，而另一種情況下，它們是單一的。但是分配輻射負載的想法是一樣的。然後接下來重要的一步，就是求平均，只有透過求平均，你才能再次得到訊號。

袁嵐峰：

Yes, so after Professor Henderson, you said that actually the molecules are not needed to be crystallized, they can just be random and you can still get the signal.

And it seems to me that if we say it can be random, we mean it is the best thing that they are totally random. If they are somewhat crystallized and somewhat of a random, it will be a disaster to us, right?

那麼在亨德森教授之後，你表示實際上分子不需要結晶，它們可以是隨機的，你仍然可以得到訊號。

在我看來，如果我們說它可以是隨機的，我們的意思是最好它們是完全隨機的，但如果他們有點結晶又有點隨機，那對我們來說就是一場災難，對吧？

約阿希姆·弗蘭克：

Yes, yes. Totally random and totally, you know, it's not good if they are only partially random that the orientations are still very much align, then you don't get information from a certain angle, that's not good.

So cryo-molecule is really, completely randomly Orient.

是的，是的。完全隨機。你知道，如果它們只是部分隨機的，方向仍然在很大程度上對齊，那麼你就不能從某些角度獲得資訊，這就不太好。

所以在這裡分子是真正完全取向的。

袁嵐峰：

So it's very interesting. So there are two extremes. It supposes that you took the extremes to us. The first one is totally ordered, the second one is totally random, but in between, that's no good, right?

這很有趣。所以這裡有兩個極端。第一個是完全有序的，第二個是完全隨機的，但在這兩者之間，就是不好的，對嗎？

And in some reviews I see that the flexibility of the molecules is not good to us, right? So it’s bond length and bond angles can change. You can change. That is not good. So we expect a very rigid molecule.

在一些綜述中，我看到分子（結構）的可變形性對我們不好，對吧？如果鍵長和鍵角是可以改變的，這就對我們不好了。所以我們期待一個非常剛性的分子。

約阿希姆·弗蘭克：

Well, I wouldn't say it's not good. It is good because that's what's supports life, the ability of the molecule to change its shape, all the properties that give rise to life have something to do with molecules interacting with one another and be able to change their conformations.

我不會說這是不好的，其實這是好的，因為這正是支撐生命的東西，分子改變形狀的能力，以及所有產生生命的其他屬性都與分子間能夠相互作用有關，以及和能夠改變它們的構型相關。

And I am always saying that X-ray crystallography does not give us a lot of information.

It gives us the information of the molecule in just one state, and that state might not be a state that the molecule assumes when it actually is doing its interactions.

我總是說X射線結晶學不能給我們提供很多資訊。

它只給我們分子在一種狀態下的資訊，而這個狀態可能不是真實情況下分子進行相互作用時的狀態。

So Cryo-electron microscopy gives us all the additional information of molecules changing their shape.

所以低溫電子顯微鏡給我們提供了分子改變形狀方面的附加資訊。

So now, but you're saying it's bad. Yes, it's bad if you cannot differentiate between the different shapes, but today's computer programs are able to actually sort the molecules out into a different one.

所以如果你說這很糟糕，那的確是的。如果你不能區分出不同的形狀那的確是糟糕的，但是今天的計算機程式實際上以及能夠將分子分類成不同的形狀了。

So if one molecule is doing this and another one is doing that, then you can by looking through hundreds of thousands of molecules. You can pick all the ones that are like this and pick all the ones that are then form averages.

所以如果一個分子某個狀態而另一個分子是另一個狀態，那麼你可以透過觀察成千上萬的分子來做實驗，你可以選擇所有的像這樣的分子，或者所有的像那樣的分子，然後平均。

袁嵐峰：

So you are partitioning molecules into subgroups.

我明白了，您是在把分子分成亞組。

約阿希姆·弗蘭克：

Yes, and we call it classification. So there are very sophisticated programs now that perform this kind of classification, and you have to realize that the raw data are very faint, single projections of molecule.

是的，我們把它稱為分類。現在有非常複雜的程式來進行這種分類，但你必須意識到原始資料是非常微弱的，是分子的單一投影。

And still these algorithms are able to sort them into populations of three -dimensional structures. So it's a really fantastic ability now that we have.

而且這些演算法仍然能夠將它們分為三維結構的種群。所以現在我們有了這種能力真是太棒了。

And I called it, story in a sample. Story in a sample, which means that, you know, you can pull all that information, all the different structures out of the same sample.

我稱之為樣本中的故事。樣本故事的意思是，你可以從同一個樣本中提取所有的資訊以及所有不同的結構。

You know you don't have to switch to another sample, but it's all of this comes out from the same sample, and then you can see two or three-dimensional constructions from the different populations and then see how they are related to each other.

你不必切換到另一個樣本，所有這些都來自同一個樣本，然後你可以看到來自不同群體的二維或三維結構，然後看它們是如何相互關聯的。

袁嵐峰：

This sounds like the basic idea of Statistical Mechanics. So it's the basic concepts of Statistical mechanics is ensemble, it's a whole lot of replicas, and you can just analyze one example and get the all the information.

這聽起來像是統計力學的基本概念。統計力學的基本概念是系綜，它包含大量的複本，你只需分析一個例子，就可以得到所有的資訊。

約阿希姆·弗蘭克：

Yeah, speaking of that, when you have a very large number of molecules, hundreds of thousand millions of them, then even states that are very lowly populated come to you. And then you can actually, you can see a continuum of states.

是的，說到這，當你有很大量的分子，比如數以億計的時候，那麼即使是布居數很低的狀態，你也能看到。然後你實際上就會看到一系列連續變化的狀態。

And when you see a continuum of states, you can use cryo-electron microscopy in order to enumerate all these different states, and they are you know, the relationship to each other is like, multi-dimensional, and then you have to sort of, you know, do some kind of dimension reduction technique to sort them, to order them in some way.

當你看到一系列連續變化的狀態，你可以用冷凍電鏡來列舉所有這些不同的分子態，它們之間的關係是類似於多維的，因此你必須做一些降維技術來把它們分類，以某種方式排列它們。

But when you have done this, then you have each element of this imagined array, you have a number of counts of a state.

Now, if you have that, then you have the energy landscape.

當你做了這些，你就得到了這個想象中的陣列的每個元素，你有一個分子態的計數。

如果你有這個，那麼你就有了能量全景圖。

袁嵐峰：

Yes, because they have the Boltzmann distribution. So we can tell the energy difference.

是的，因為它們符合玻爾茲曼分佈。所以我們可以分辨出能量的差別。

約阿希姆·弗蘭克：

Exactly, so we are now able to reconstruct the energy landscape of molecule. Yeah, so there's just been an article that we published in Nature Communications.

沒錯，所以我們現在可以重建分子的能量全景圖了。【袁：這真是太美妙了！】這正我們剛剛在《自然通訊》上發表的一篇文章的內容。

袁嵐峰：

That's great. So is that the benefit of the cryo-EM is the better the XRD right? XRD cannot do this?

太棒了。那麼這是冷凍電鏡比較於X射線衍射的一個優點嗎？X射線衍射不能做這個，對吧？

約阿希姆·弗蘭克：

No, it cannot, yeah.

是的，它不能。

擴充套件閱讀：

袁嵐峰對話諾獎得主安德烈·蓋姆（上）用膠帶撕出石墨烯，諾獎得主：我們的想象不該被限制

袁嵐峰對話諾獎得主安德烈·蓋姆（下）對俄羅斯感情複雜，中國應用科學世界第一

諾獎得主蓋姆為什麼認為中國的基礎研究僅次於美國？奈米限域毛細凝聚可以作為一個例子｜袁嵐峰

袁嵐峰對話諾獎得主約阿希姆·弗蘭克（一）神器冷凍電鏡究竟是什麼？諾獎發明者親自解說 | 科技袁人

作者簡介：袁嵐峰，中國科學技術大學化學博士，中國科學技術大學合肥微尺度物質科學國家研究中心副研究員，科技與戰略風雲學會會長，“科技袁人”節目主講人，安徽省科學技術協會常務委員，中國青少年新媒體協會常務理事，入選“典贊·2018科普中國”十大科學傳播人物，微博@中科大胡不歸，知乎@袁嵐峰（https://www.zhihu.com/people/yuan-lan-feng-8）。

背景簡介：影片釋出於2020年12月14日《冷凍電鏡：只要我凍得足夠快，分子就以為自己在家裡》（https://www.ixigua.com/6906033244232221191）。